Published online 24 September 2007

Nucleic Acids Research, 2007, Vol. 35, No. 19 e126 doi:10.1093/nar/gkm559

Cre reconstitution allows for DNA recombination selectively in dual-marker-expressing cells in transgenic mice Yanwen Xu1, Gang Xu2, Bindong Liu3 and Guoqiang Gu1,* 1

Program in Developmental Biology and Department of Cell and Developmental Biology, Vanderbilt University Medical Center, Nashville, TN 37232, USA, 2Department of Medicine and Therapeutics, Prince of Wales Hospital, The Chinese University of Hong Kong, Shatin, Hong Kong and 3Center for AIDS Health Disparities Research, Meharry Medical College, Nashville, TN 37208, USA

Received May 2, 2007; Revised July 8, 2007; Accepted July 9, 2007

ABSTRACT Cre/LoxP-based DNA recombination has been used to introduce desired DNA rearrangements in various organisms, having for example, greatly assisted genetic analyses in mice. For most applications, single gene promoters are used to drive Cre production for conditional gene activation/inactivation or lineage-tracing experiments. Such a manipulation introduces Cre in all cells in which the utilized promoter is active. To overcome the limited selectivity of single promoters for cell-type-specific recombination, we have explored the ‘dual promoter combinatorial control’ of Cre activity, so that Cre activity could be restricted to cells that express dual protein markers. We efficiently reconstituted Cre activity from two modified, inactive Cre fragments. Cre re-association was greatly enhanced by fusing the Cre fragments separately to peptides that can form a tight antiparallel leucine zipper. The co-expressed Cre fusion fragments showed substantial activity in cultured cells. As proof of principle of the utility of this technique in vivo for manipulating genes specifically in dual-markerpositive cells, we expressed each inactive Cre fragments in transgenic mice via individual promoters. Result showed the effective reconstitution of Cre activates LoxP recombination in the co-expressing cells. INTRODUCTION The Cre/LoxP system utilizes P1 bacteriophage Cre recombinase to catalyze recombination between tandem LoxP DNA sequences (1,2). This system has been widely used in multiple organisms, including yeast (3), plants

(4–7) and animals (8–14). The Cre/LoxP technology is particularly useful for mammalian genetics, because it allows the analyses of essential genes in specific organs by gene inactivation (8–15) or controlled ectopic gene expression (16,17). When combined with visible marker proteins, Cre-LoxP-based gene activation allows for cell marking and cell lineage analyses in living animals (17). Specific gene promoters are usually utilized to drive Cre expression in desired tissues. Thus, the promoter specificity limits where Cre can be expressed. To this end, most available promoters drive gene expression in multiple cell types. This deficiency has greatly limited our ability to manipulate genes within specific cells, such as stem cells that can only be identified by their expression of several molecular markers (18–20). An approach that introduces Cre exclusively to cells that express more than one protein marker would facilitate our understanding of the function and fate of specific cells in vivo. Active protein can be reconstituted from peptide fragments of corresponding molecules. For some proteins, fragmented peptides can directly re-associate to restore activity (21–23). In other scenarios, assisted protein reconstitution is required. In this latter case, protein can be cleaved to two inactive fragments. Each fragment was then fused to one of a pair of interacting protein motifs respectively. The interacting motifs could bring the protein fragments to proximity to facilitate efficient reassembly (23–30). Both the above schemes have been explored for Cre activity reconstitution (31,32). In one report, cDNA molecules were designed to produce two inactive Cre halves in same cells. This approach, combined with improving protein translation from Cre cDNA (33), was reported to reconstitute 32.5% of wild-type Cre activity (32). In the other case, cDNA molecules were constructed to produce two inactive Cre moieties connected with FK506-binding protein (FKBP12) and FKBP12rapmycin-associated-protein (FRP), respectively. Because the interaction between FKBP12 and FRP was FK506

*To whom correspondence should be addressed. Tel: +1 615 936 3634; Fax: +1 615 936 5673; Email: [email protected] ß 2007 The Author(s) This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/ by-nc/2.0/uk/) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.

PAGE 2 OF 9

e126 Nucleic Acids Research, 2007, Vol. 35, No. 19

dependent, Cre activity could be restored only when both moieties and FK506 were present (31,34). This method restores
3–4% Cre activity (31). The usefulness of these two systems in animal models has not been reported. We attempted to reconstitute Cre in mouse cells that express two protein markers. Because we could obtain less than 2% Cre activity using the published Cre fragment complementation process (32), we utilized assisted-Cre reassembly for this purpose. The Cre open reading frame (ORF) was cleaved into two cDNA fragments, each encoding an inactive Cre peptide. Each cDNA fragment was then fused to an ORF for one of two peptides that could form antiparallel leucine zippers (35). This leucine zipper was artificially designed and has been reported to effectively assist protein reconstitution in vitro and in vivo, and these peptides do not seem to interfere with normal cellular functions (25,28,35). When these two modified Cre fragments were co-expressed in tissue culture,
30% Cre activity could be restored, an 8-fold improvement over previously published methods (31,32). When expressed in the pancreatic tissue of transgenic mice from individual promoters, the inactive Cre fragments effectively induce LoxP-based recombination. This approach opens the possibility to study gene function or perform lineage labeling in cells that express dual protein markers in animal models.

Figure 1. A diagram of the half-Cre molecules and the interacting peptide sequences. (A) The Cre molecule was designed to be cleaved into two molecules between two glycine residues (amino acid residues 190–191, as numbered in X03453). The N-terminal half was fused with one of a pair peptides that form a leucine zipper (N-peptide), whereas the C-terminal half was fused with the other peptide (C-peptide). (B) The C- and N-peptide sequences.

MATERIALS AND METHODS

Table 1. DNA oligos sequence utilized in this report

DNA constructs and transgenic mouse production

Primer name

Primer sequence

T3 T5 nX5 N3 Cz1 Cz2 Cz3 Nlc

gtcgaccaccatggttaaagatatctcacgtactg gcggccgctcaaatatggattaacattctcccac actcgagccaccatggcacccaagaagaagaggaaggtgtcc gcggccgcctaatcgccatcttccagcaggc actcgagaccaccatggtgtccgaacaactggagaagaagctcca ggctctcgaaaagaagctggctcaactcgaatggaagaatcaagc ctcgaatggaagaatcaagctctggaaaagaagctcgcccaaggctc actcgagaccaccatggtgcccaagaagaagaggaaggtgtccga acaactggagaaga gaaaagaagctcgcccaaggctctggtgggagaatgttaatccatatt tgcggccgctcattgagccagctccttcttcagagcttggagttcccact ttcagagcttggagttcccacttcagctgagccagttccttcttattagcttg gccagttccttcttattagcttggagctccttcttcagggcaccag ctccttcttcagggcaccagaaccaccgtcagtacgtgagatatc tagcttgggatctttgtgaaggaaccttacttct ggcggccgcagatcgatccagacatgataagatac ctcgagccaccatggcacccaagaagaagaggaaggtg ccttcacaaagatcccaagctaga tggttatgcggcggatccgaaaa tccgtctctggtgtagctgatga gcggccgcagatctaatcgccatcttcca ctcgagccaccatgtccaatttactgaccgtacac gcggccgcttacaccttccttttctttttcggaccatcgccatcttccagcag

For cDNA encoding the fusion of leucine zipper-forming peptides with Cre moieties (Figure 1), overlapping DNA oligos were synthesized and PCR-amplified with nlsCre cDNA [with a nuclear localization signal (NLS) present in Cre’s n-terminus] as template (36). One final cDNA ORF (called nCre) encodes a protein with N-terminal half of Cre fused with N-peptide at Cre C-terminus (with a NLS at its n-terminus, Figure 1). The oligos utilized were: X5, Nz1, Nz2, Nz3 and nZ4 (Table 1). Anther cDNA ORF (called cCre) encodes a protein with the C-terminal half of Cre fused with C-peptide at Cre n-terminus (Figure 1). The DNA oligos used were: N3, Cz1, cZ2, cZ3 and cZ4 (Table 1). To add an extra nuclear localization signal coding sequence to cCre in its 50 end (to produce nlcCre), we utilized the following oligos: Nlc, N3, Cz1, cZ2, cZ3 and cZ4 (Table 1). PCR fragments were cloned into the pBluescriptKSII vector to produce pYW415, pYW429 and pYW418, respectively. The XhoI-NotI fragments from these constructs were ligated into the corresponding sites of the pCIG-expression vector, containing the CMVchicken-b actin promoter to drive gene expression, to produce pYW427, pYW443 and pYW425 (37). For CMVstop-GFP, an EcoRI–SpeI fragment (contains a Poly A signal) from pBS302 (38) was ligated into the EcoRI–SpeI sites of pGreenlatern-1 to produce pYW421 (39). As control for Cre activity assay, the full-length Cre (which was PCR amplified and inserted into the XhoI–NotI sites of the pCIG vector to produce pYW482. The oligos utilized were: fc1 and fc2 (Table 1). In order to use human ubiquitin promoter (Ubc) to drive nlcCre expression,

Cz4 Nz1 Nz2 Nz3 Nz4 PA1 PA2 Fc1 cga Nz1 Cz1 Fc2 X5 nlsb

the SalI (fill-in)–NcoI fragment from pYW418 was cloned into the NcoI–NotI (fill-in) site of Ui4-GFP-SIBR vector (40). Note all reading frames contain an idealized ‘Kozak sequence’ CCACC before ATG. To amplify the a5, b1, b1-nls fragments reported in (32), DNA oligos (X5+T5), (N3+T3) and (N3+nlsb) were utilized. The pCIG vector was utilized to drive the expression of these fragments as well.

PAGE 3 OF 9

For transgene constructs, PCR-amplified SV40 polyA sequences from pGreenlatern1 were inserted into the SmaI site of pBluescript KSII, producing pGD103 (oligonucleotides utilized: pA1, pA2; Table 1). The XhoI–NotI (filledin)-digested nCre or nlcCre fragments were inserted into the XhoI–EcoRV site of pGD103, producing YW452 and YW451, respectively. Finally, XhoI (filled-in)–NotI fragments from YW452 and YW451 were inserted downstream of the murine Pdx1 promoter (SmaI/NotIrestricted plasmid #571; gift from C. Wright). Inserts were released with SalI–NotI for transgenic animal production in the Shared ES Cell/Transgenic Animal Resources in Vanderbilt Medical Center. nCretg and nlcCretg genotyping was with ng1 and cga oligonucleotides (Table 1). R26R-EYFP (R26YFP) and Z/AP reporter animals genotyping, and alkaline phosphatase detection was by published methods (36,41). All mouse care, handling and crosses followed IACUC protocol M/03/ 354 (Gu), approved by the Animal Welfare Committee of Vanderbilt Medical Center. Cre activity assay A reporter plasmid (YW421) expressing eGFP Cre dependently was used to assay Cre activity. Reporter YW421, plus cCre or nCre plasmids (or both), and mCherry-producing plasmid (42) were co-transfected into HK293 cells using calcium-phosphate-based technique. After 14–16 h, transfected cells were analyzed by flow cytometry for fluorescence expression. The percentage of red cells that express eGFP was plotted against the Cre plasmid(s) concentration. For most transfections, 0.2 mg YW421, 0.1 mg mCherry and 0.1–2 ng Cre-expressing plasmids were utilized for each well of 12-well dishes. With bigger wells, the plasmid amount was scaled-up proportional to the well area. To ensure that the Cre activity comparisons were made in a linear range, a standard curve was constructed varying the concentration of Cre-producing plasmid, measuring the output green/red ratio. Reconstituted Cre activity was calculated against this standard curve. All assays used a minimum of triplicate samples. All assays utilized nlsCre as a control. Immunofluorescence/immunohistochemistry Established protocols were used. Briefly, tissues were fixed in 4% paraformaldehyde overnight at 48C, or 4 h at room temperature, and prepared as frozen sections. Frozen sections conserve GFP fluorescence. Primary antibodies used were: guinea pig anti-insulin and guinea pig antiglucagon (Dako, Carpinteria, CA, USA); rabbit anti-SS, guinea pig anti-PP (In Vitrogen, Carlsbad, CA, USA); rabbit anti-amylase, biotinylated Dolichos biflorus agglutinin (DBA, Sigma, St Louis, MO, USA). Secondary antibodies used were: Cy3-conjugated donkey anti-rabbit IgG, Cy3-conjugated donkey anti-guinea pig IgG, Cy3-conjugated streptavidin (Jackson Immunoresearch, West Grove, PA, USA). Alkaline phosphatase staining followed reported protocols (43). All antibodies utilized 1:1000 dilutions.

Nucleic Acids Research, 2007, Vol. 35, No. 19 e126

Microscopy For whole mount YFP fluorescence, tissues were dissected and fixed in 4% paraformaldehyde (overnight, 48C), washed and mounted in PBS in chambers on glass slides. Samples were observed using either inverted fluorescence microscope (for regular observations) or confocal microscopy (for high quality pictures). Confocal imaging was also utilized to observe immunofluorescence-stained samples. Typically, 0.4 mm optical z-sections were taken for thick samples. A maximum of two adjacent optical sections were stacked and projected to produce a high quality picture for each figure.

RESULTS Creating inactive Cre fragments for reconstitution Reconstituting Cre activity from two inactive peptide fragments could benefit from a pair of interacting protein motifs to bring Cre fragments to proximity for refolding. Additionally, Cre ORF needs to be cleaved at a specific site so that the encoded Cre fragments will be inactive, yet are able to reassemble into an active molecule when brought together. We considered several criteria in choosing interacting protein motifs to assist in Cre reconstitution, including high affinity, high specificity, and lack of dominantnegative effects in living cells. The reported pair of antiparallel, heterodimer leucine zipper-forming peptides (Figure 1, named as N- and C-peptide) fit this profile (44,45). These peptides were shown to effectively assist protein folding both in vitro and in vivo with no detectable dominant negative effects in living animals (24,25,44,45). To choose the best point to separate Cre into two portions, we examined the Cre 3D structure for the residues and secondary structures that are crucial for its activity (4,32,46). We choose to generate open reading frames (ORF) which encodes Cre amino acid residues 1–190 and 191–343, separated between two glycine residues at 190 and 191 (Figure 1). The flexibility of the peptide bond between glycine and other amino acid residues makes it more likely to tolerate addition of extra peptides without disrupting the secondary and tertiary structure of Cre. In addition, these two glycine residues are localized between two b-sheets and are expected to point away from the DNA elements during recombination (46). Therefore, connecting leucine zippers with each half of the Cre protein at this position is expected to minimally interfere with Cre function. We derived three cDNA ORFs that encode three Cre fragments, nCre, cCre and nlcCre, for Cre reconstitution (Figure 1A). nCre was a fusion between the N-peptide to the N-terminal half Cre. We included the SV40 large T antigen NLS in the N-terminus of this molecule. cCre was a fusion between the C-peptide to the C-terminal half Cre (Figure 1B). nlcCre also contains a SV40 large T antigen NLS in its n-terminus, otherwise, it is identical to cCre (Figure 1). The presence of an NLS in both N- and C-terminal half Cre molecules respectively is likely to restrict both molecules in the nucleus and allow for

PAGE 4 OF 9

e126 Nucleic Acids Research, 2007, Vol. 35, No. 19

efficient interaction. In order to express these expected protein fragments, the ORFs were put under the control of the CMV-b actin promoter in the pCIG vector. Leucine zipper-forming peptides in Cre fragments assist Cre reconstitution The substrate for testing reconstituted Cre activity was a reporter plasmid that produces eGFP in a Cre-dependent manner (Figure 2A). The reporter plasmid was transfected into HEK293 cells in large excess (see Materials and methods section), together with Cre-producing plasmids. An mCherry-producing plasmid (42) was co-transfected as a control for cell transfection efficiency, with the green/red fluorescence ratio providing an index for Cre activity. When 0.1–1 ng Cre-producing plasmid was used per well (6-well dishes), the green/red fluorescence ratio versus [Cre] was linearly correlated (Figure 2B–D), demonstrating that this method could be used for Cre activity assay. With this activity assay, we carefully compared the Cre activity restoration using unassisted and assisted reconstitution approaches. We generated three ORFs that encode the a5, b1 and b1nls peptides reported in Casanova et al. (32) and put them under the control of the pCIG promoter. These fragments did not have interaction motifs, yet were reported to reconstitute substantial Cre activity. Using our established assay, we found that both (a5 + b1) and (a5 + b1nls) restored